-
Page 1 of 7
The benefits of microsimulation for large area models
Author :
Pete Sykes, Iain Clement, SIAS Limited Nick Cohn, Niels van den
Brink, Grontmij
ABSTRACT
There is a perceived wisdom amongst traffic modellers that a
principal criterion for choice of modelling tool is the size of the
road network, and that microsimulation is only suitable for small
geographical areas. We challenge this notion, and demonstrate the
benefits of using microsimulation for large models. By deploying
microsimulation over a wide area, study boundaries can be defined
with the aid of the model itself, and the problem of identifying
the sometimes arbitrary limits to the modelled road network are
diminished.
Introduction
Microsimulation is a well known technique for road traffic
simulation in which the actions of every individual vehicle are
evaluated at sub second intervals. Their interactions with the road
network, the control systems and with the other vehicles on the
network are determined and hence their journeys through the road
network are reproduced. Thereafter when changes in the simulated
road network or the control system are made, their effect can be
determined and a benefit analysis undertaken.
To many transport professionals, the prefix micro in
microsimulation is assumed to apply to the size of area that can be
modelled not to the detail of the interactions. This overloading of
the term leads them to assume that microsimulation can only be
applied to small areas and some other technique must be used to
model larger areas. While it is true that initially microsimulation
was used for single junctions and the optimisation of traffic
signals at a local level, in recent years, the combination of
faster computers and new software developments has made it possible
to apply microsimulation at a regional level. These regional
microsimulation models are now being applied to answer questions
that until recently were studied using static or semi-dynamic
models. While microsimulation models do not yet typically contain
the components of trip generation, distribution or mode choice,
they provide added value at the regional level for testing traffic
management strategies and detailed dynamic measures.
As our road network becomes more and more complex, the added
value of standalone traffic management measures is no longer
sufficient. Intelligent and dynamic measures applied at the network
level are more often required for strategic traffic management.
Microsimulation is now an important tool for testing combinations
of, for example, ramp metering, dynamic route information and
intelligent signalised junctions (such as SCOOT or Utopia Spot).
However drivers are fundamentally selfish beings who act to
maximise their benefit from changes in the road scheme. They learn
of the effect of changes and change their behaviour accordingly.
Hence implementing a junction improvement or an ITS system may
reduce congestion in that area and drivers who previously had
avoided that junction will now use it if it improves their journey
time. The knock on effects can be wide ranging and the overall
benefits may not be clear if only the area to be changed is
modelled in detail.
Here we discuss the capabilities and the benefits of using
microsimulation for the entire task rather than adopting a hybrid
approach. We do this based on two specific case studies: the model
of the Alkmaar region in the Netherlands and a study undertaken for
the Scottish Executive on the provision of overtaking sections on
single carriageway roads. The two case studies demonstrate
technical as well as organisational elements of applying regional
microsimulation models. The paper covers the most
-
Page 2 of 7
important steps in model development, validation and
application, such as the data requirements, calibration methods,
validation, application for traffic management strategies and a
comparison with static regional models.
Alkmaar Region Using the simulation model as a planning tool
In the Netherlands, the responsibility for planning and
performing road maintenance as well as new road infrastructure is
divided among national, provincial and municipal organisations.
This division is based upon road classification. Each organisation
creates its own roadworks program and because roadworks inevitably
influence traffic conditions, an effort is made to coordinate the
different programs within a given region.
In addition to roadworks, planning for and taking account of
major events requires planning at a regional level to ensure
accessibility and anticipate traffic problems. Ideally, this
planning is done at the level of decision makers, practitioners and
the public in order to optimise regional accessibility by way of
joint planning, joint decision making and joint communication.
In the Alkmaar, North Holland region, the different
organisations concluded that an automated tool would be useful for
centrally coordinating the large number of projects and events. By
visualising cumulative effects of projects and events, the
authorities could gain insight into the interaction effects of
different combinations of projects and could consequently make
informed decisions. The decision was made to use a microsimulation
model because of the capabilities for simulating temporary road
conditions in a realistic manner, such as lane closures, detours
and temporary situations at junctions. The ability to record
vehicle-specific travel times was useful for establishing police
and fire service travel times during construction and maintenance
periods.
Model area
Figure 1
Figure 1 shows the area to be modelled and the corresponding
S-Paramics road network. Figure 2 shows one example of a junction
in the model
-
Page 3 of 7
Figure 2
The simulation model was developed based on the conversion of an
existing static model of Alkmaar. This meant that the core area of
the model could be created efficiently and that network
characteristics and matrices did not have to be recreated for a
large portion of the model. Detailed traffic signals and road
geometry were added to complete the model and all the traffic
signals in the model were included as vehicle-activated signals.
This was essential for a realistic simulation but also ensures that
the available capacity of the road system is being used
optimally.
With S-Paramics it is possible to include different simulation
periods within a single model. In this case, the morning and
evening peak periods, the evening period, the interpeak period and
a weekend period were modelled. This was important because the
interpeak, evening and weekend periods contain extra available
capacity that can be used for road works. The model results provide
insight into the options for the time of day or day of week for
carrying out specific works.
The model has added value for the development of an optimal road
works and phasing plan. The visualisation of the traffic situation
for alternative plans is of particular importance to the involved
organisations for establishing agreed priorities and making joint
choices. The quantitative reporting is then required for testing
and demonstrating that the choices result in the best possible
accessibility in the region.
The model was validated for the current situation (2005), prior
to the beginning of various major road works in the region. This
was completed using available traffic counts and the Paramics
matrix estimation facility. During this process, it was essential
that the organisations involved were satisfied with the picture of
regional and local traffic situations in the modelled periods and
agreed that it was a realistic representation of the actual road
system. Without trust in the model, the joint works planning
process would have been much more difficult to achieve.
The simulation model provides detailed insight into the effects
of road works and events. The forecast effects and the duration of
road works are retained in a database. This data includes the
rerouteing observed in the model, the initial mitigation measures
planned for the event and any extra measures found to be required
during testing. The database is a living planning instrument and is
continually updated and expanded.
Alkmaar Region N242 Provincial Highway
The largest road works project in the Alkmaar region is the
reconstruction of the provincial highway N242. With this specific
project in mind, a regional accessibility index was developed to
aid in establishing the best works phasing and planning. This index
was expanded for use with all projects in the region. It consists
of the following elements:
-
Page 4 of 7
Travel times for emergency services
Public transport delays, based on specific, trips and acceptable
travel times
The accessibility of office and industrial parks and the centres
of regional towns
Traffic conditions, determined by travel times on the Alkmaar
ring and other major roads
The planning of the mitigation measures for all works and events
is done in detail. It can include lowered speed limits, narrow or
restricted lanes, turning restrictions and alternate route
guidance. All of these measures may be implemented by hour and each
temporary situation can have a significant effect on traffic
conditions and on journey times.
All the relevant organisations were requested to submit their
plans for road works and events and based on this information the
accessibility index was evaluated. An inventory of 100 projects
between 2005 and 2008 was collated and the planning group was
requested to determine, using the simulation model, which works
could be carried out concurrently. Optimal timings were found
either by adjusting the hourly mitigation measures or by
rescheduling the work to avoid combinations of events that caused
excessive degradation to the accessibility index.
The use of the Paramics microsimulation model in the N242 works
planning process showed that the iterative testing and
visualisation of alternative planning and phasing works well in
developing consensus among authorities and with the public. The
accessibility index, which can be tailored to local requirements,
makes it possible to evaluate design-and-construct proposals in an
impartial way that is directly related to the quality and goals of
a works project.
The regional scale model was necessary in order to include
possible interaction effects between different road works and/or
different mitigation measures. The detail in planning with the
options of using ITS systems for dynamic speed control, lane
restrictions and route guidance made the use of a microsimulation
model essential and this had to be achieved on a regional scale to
allow the effects of each change to properly propagate to their
natural extent.
Scotland: Using the simulation for economic assessment of
overtaking sections
On a single carriageway, a number of vehicles may be slowed by a
single slow lead vehicle, typically an HGV or an agricultural
vehicle. Consequently, vehicles will form a platoon behind the lead
vehicle for as long as they are unable to pass it. If an
improvement scheme includes carriageway widening to facilitate
dedicated or opportunistic overtaking, then the benefit of the
scheme will largely be felt by the vehicles that can now pass the
lead vehicle, known as platoon dispersion. It is important to note
that the benefit is not felt just over the length of the scheme but
for some distance downstream. Microsimulation can capture this
detailed effect and PEARS (Programme for Economic Assessment of
Road Schemes) can translate it into a monetary value to be included
in the overall cost benefit analysis.
S-Paramics includes an overtaking model in which a vehicle will
first assess its desire to overtake based on its target link speed
and the achieved speed of the vehicle ahead. It will then assess
its ability to overtake based on the gap available ahead of the
vehicle to be overtaken and also the visibility of the road ahead
and presence of an oncoming vehicle in that space. The manoeuvre
will be initiated when both desire and gaps are present.
A recent commission for the Scottish Executive, on behalf of all
UK government transportation departments, was designed to assist
with the preparation of a new technical advice note for wide single
2+1 overtaking lanes looked at a number of parameters,
including::
economic scheme length
extent of downstream benefit
platoon lengths
-
Page 5 of 7
In all cases, the benefits were compared to a reference base of
either standard S2 (7.3m) or Wide Single WS2 (10.0m) carriageway.
Wide Single WS2+1 carriageways comprise the development of a third
traffic lane to provide dedicated overtaking. Traffic in the
opposing direction is prohibited from overtaking.
Further benefits can be accrued through provision of alternating
overtaking lanes arranged in either conflicting or non-conflicting
configurations. In conflicting, or critical, configurations, the
direction of alternate overtaking is towards the mid point of the
scheme (nose to nose). In non-conflicting, or non-critical,
configurations, the direction of alternate overtaking is towards
the ends of the scheme (tail to tail). The differences to the
driver lie in the changes in platooning levels, and hence gaps, in
the oncoming traffic as the overtaking manoeuvre is to be attempted
A simulation model was developed to replicate a 15km section of
single carriageway. The model was coded to include three sections;
an approach length, a central study section followed by a run-out
section. The approach, or lead-in, was around 4.5km in length in
which overtaking was alternately allowed and barred, this ensured
that realistic platoons had formed at the start of the study
section. The study section varied in length according to the each
test. The run-out, or downstream section, of 10km allowed a measure
of downstream benefits.
The model was calibrated by adjusting the overtaking parameter
so that the vehicle order change in the model matched that from a
set of number plate matching surveys. In all models, the HGV
proportion was fixed at 10% and both conflicting and
non-conflicting overtaking lane arrangements were evaluated.
With the calibrated models developed, various test scenarios
were replicated. For each scenario, an equivalent Do-Minimum
scenario was also developed, against which the scheme was assessed.
Both were modelled for opening year and design years, although it
should be noted that PEARS can deal with multiple assessment years.
Each Do-Minimum and Do-Something scenario was simulated 10 times
using a random seed value to ensure a distribution of results. The
assessment was then repeated for three or four different flow
ranges.
The first assessment looked at economic justification for
various scheme lengths. The test model assumed that the road would
be upgraded from either a standard 7.3m S2 single carriageway or
10.0m WS2 wide single carriageway to 12.5m WS2+1 carriageway. In
all models, the HGV proportion was fixed at 10% with the flow range
varied between 5,000 and 15,000AADT.
S2 / WS2+1 (Non-Conflicting) - Economic Justification10%
HGVs
-0.50
-0.40
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
0 500 1000 1500 2000 2500 3000
Scheme Length(m)
Net
Pre
sent
Val
ue (N
PV)
(M
illio
ns /
100m
)
5000 AADT 10000 AADT 15000 AADT
Figure 3 : S2 to WS2 Economic analysis
-
Page 6 of 7
The results of the S2 to WS2+1 assessment are summarised in
Figure 3.
The results of the assessments indicated that:
S2 to WS2+1 would be difficult to justify economically in low
flow situations (
-
Page 7 of 7
that the road would be upgraded from either a standard 7.3m S2
single carriageway or 10.0m WS2 wide single carriageway to 12.5m
WS2+1 carriageway. Again, in all models the HGV proportion was
fixed at 10%. The results of the S2 to WS2+1 assessment are
summarised below in Figure 5.
Figure 5 : S2 to WS2 Platoon Lengths
The analysis of the platoon profile exiting the study section
clearly indicted a reduction in the number of longer platoons in
favour of more shorter platoons of around 4 or 5 vehicles. The
assessment concluded:
number of vehicles in platoons >5 reduces from 60% to 48% of
the total number of vehicles
number of vehicles in platoons >10 reduces from 33% to 23% of
the total number of vehicles
By evaluating the economic benefits in detail using a
microsimulation model it is possible to capture very small changes
in travel time and vehicle operating costs which would otherwise be
lost. A cost benefit package such as PEARS then enables these
savings to be translated in to monetary benefits and the overall
cost benefit analysis of a proposed road improvement scheme made
more accurately. However the main benefit is felt not in the area
where the work is undertaken, it is felt for many kilometres
downstream. Simply modelling the immediate area where the work is
to be done will not capture the true effects of the work and
modelling the wider area with macro level assignment techniques
will not capture the detailed but significant effect of the
formation and dispersion of platoons of vehicles.
Conclusion
Both of these examples show the benefits of using
microsimulation in the wider area in one homogenous model with
sufficient detail to undertake the necessary analysis. Simple
assignment methodologies cannot capture detailed effects and the
benefits of a road improvement scheme are frequently due to the
accumulation of just these effects. By drawing a line around where
we use microsimulation and where we use assignment methods we are
in effect stating that we know the limits of the effects of the
changes which presumes the output of the modelling process. As
shown in overtaking study this is not necessarily true.
S 2 / W S 2 + 1 P la t o o n L e n g t h s : O U T P r o f ile1
0 ,0 0 0 A A D T , 1 0 % H G V s
0
1 0 0
2 0 0
3 0 0
4 0 0
5 0 0
6 0 0
7 0 0
2 3 4 5 6 7 8 9 1 0
P la t o o n L e n g t h ( V e h ic l e s )
Num
ber o
f Veh
icle
s
S 2 O U T W S 2 + 1 O U T
